Comments

Levels corrections are the easiest way to get decent color quickly. It is a correction you can make almost without checking the image visually first (although it is always a good idea to make a visual check of any image you are correcting). The exceptions to going ahead with Levels corrections as described are for images that you know have desired color casts (such as sunset or sunrise pictures, in which there can often be a decided and necessary color shift toward yellow and red). Doing this correction to images with color casts you want to keep — or those which are too extreme — will probably ruin the desired tone and color of the image. While this Levels technique often corrects quite nicely for fluorescent and most other man-made lighting flavors, it is not an excuse to use bad lighting.

Histograms with long tails, or images in which one or two of the three colors seems extremely compressed, may suggest that there is some intentional distortion, such as the use of a filter by the photographer. It also may suggest that you have a bad scan or other digital representation. Some images are meant to look light, and some are meant to look dark: when changing the Levels try to keep the changes in character with the image (i.e., high-key images may look their best if adjustments keep them high key rather than attempting to darken the image).

The more extreme the changes you make in the Levels correction, the more chance you will have to damage the image. If you have a long tail to correct, do it a little at a time (make two or three separate corrections, closing out of the Levels in between). Some experience doing these corrections will help you know what to do, when to use more than one correction, or when to leave well enough alone. Correcting for a lengthy tail in any histogram may produce poor results if you choose to cut off the tail completely. As a general rule, the longer the tail, the less likely it is that you should cut it off; cut half or a quarter off if the bulk of information in the channel seems compressed. If the information in the channel is very compressed (with the bulk of the information falling in one-third or less of the potential range), meet the tail only partway. The middle ground is a better way to go in corrections, especially if only one of the channels has such a tail.

If all channel histograms look similar (each has a similar tail), do approximately the same thing to each of the channels (if you cut the tail on one, cut the tail on all). If all look very different, you will have to make a judgment call, based on visual examination of the image. Experience helps with this type of evaluation and correction, so the more you do, the easier it will become.

 

  Copyright © Richard Lynch 2001